Role of meteorological controls on interannual variations in wet‐period characteristics of wetlands
Role of meteorological controls on interannual variations in wet‐period characteristics of wetlands
复制标题
DOI：
10.1002/2015wr018493
复制
发表时间：
2016-07
影响因子：
5.4
通讯作者：
Yanlan Liu;Mukesh Kumar
中科院分区：
文献类型：
--
作者：
Yanlan Liu;Mukesh Kumar
湿地的许多生态功能都受到湿期的影响，即，地下水位（GWT）持续接近地表的时间间隔。因此，有一个至关重要的需要，以了解控制年际变化的湿期。鉴于湿地GWT长期测量的稀缺性，仅使用测量方法来了解湿期的变化具有挑战性。在这里，我们使用了一个基于物理的，完全分布式的水文模型，与公开的水文数据协同作用，模拟在美国东南部流域的10个内陆森林湿地的长期湿期变化。然后实施贝叶斯回归和变量选择框架，以（a）评估模拟的湿润期可以通过降水量（Ppt）和潜在蒸散量（PET）估计和预测的程度，以及（B）推断季节性Ppt和PET的相对作用。我们的结果表明，在32年的模拟期间，湿期开始日期和持续时间可能会变化超过6个月。值得注意的是，60-90%的这些变化可以捕获使用回归的基础上，在大多数湿地的季节性PPT和PET。季节气象条件对丰水期变化的影响是不均匀的，这表明年变量可能无法解释丰水期的年际变化。贝叶斯框架能够预测湿期变化，误差小于1个月，置信水平为90%。所提出的框架提供了一个最小化的方法来估计和预测湿地的湿期变化，并可用于了解未来的湿地相关生态功能的响应。
Many ecological functions of wetlands are influenced by wet‐periods, i.e., the time interval when groundwater table (GWT) is continuously near the land surface. Hence, there is a crucial need to understand the controls on interannual variations of wet‐periods. Given the scarcity of long‐term measurements of GWT in wetlands, understanding variations in wet‐periods using a measurement approach alone is challenging. Here we used a physically based, fully distributed hydrologic model, in synergy with publicly available hydrologic data, to simulate long‐term wet‐period variations in 10 inland forested wetlands in a southeastern US watershed. A Bayesian regression and variable selection framework was then implemented to (a) evaluate the extent to which the simulated wet‐periods can be estimated and predicted by precipitation (Ppt) and potential evapotranspiration (PET) and (b) infer the relative roles of seasonal Ppt and PET. Our results indicate that wet‐period start date and duration could vary by more than 6 months during the 32 year simulation period. Remarkably, 60–90% of these variations could be captured using regressions based on seasonal Ppt and PET in most wetlands. Effects of seasonal meteorological conditions on wet‐period variations were found to be nonuniform, which indicate that the annual variables may not explain interannual variations in wet‐periods. The Bayesian framework was able to predict wet‐period variations with errors smaller than 1 month at a 90% confidence level. The presented framework provides a minimalistic approach for estimating and predicting wet‐period variations in wetlands and may be used to understand the future responses of associated ecological functions in wetlands.